Soft-computing approach to the design of current mirror circuits

This paper describes a method for the layout design of current mirror circuits. The method utilizes the soft computing approach. Specifically, a genetic algorithm (GA) performs placement and routing of the current mirrors. Crossover deals with the spread out of best routing strategies over the population, while mutation deals with better placement and improvement of routing strategies. Each run of the GA produces new routing strategies expressed in terms of fuzzy IF-THEN rules. Rules are tuned so that search procedure becomes less and less random. As a result, better tractability of the searching problem is achieved